Scheduling methods for large groups

[PWTom]

I just inherited our Pack's Pinewood Chair and I am looking for an improved racing schedule. We have a 6-lane aluminum track with electronics to determine the order of finish in a heat. Upwards of 70 to 80 boys participate.

In previous years, we have used the Stearns charts for either 72 or 84 cars. Trophies are awarded to the fastest three cars in the Pack overall and medals awarded to the fastest three cars in each rank. Based on my experience, and some reading among the various forums, I think the Stearns method is sufficiently inaccurate for awarding the top three trophies. I would like to run some method to select the six fastest cars from the entire pack to move to a finals round, then use another chart to determine the racing and outcome of the finals.

My questions are what scheduling methods would you suggest for handling the large, full pack racing schedule to determine which cars move into the finals, and what schedule for the finals?? Your comments would be greatly appreciated. Thanks!!

[Stan Pope]

Reply 1 of several ...

Your suspicion about the pack's past method is well founded. The probability that the fastest three cars are among the top three finishers (not necessarily ranked correctly) is disgusting! The probability that the fastest three within a subgroup such as Tigers are among the three highest scoring members of that subgroup is worse.

The problem lies in equality of opposition (or, rather, lack of equality of opposition.)

The fact that about 80% of each car's opponents are in another group is a contributing complication to accuracy.

Accuracy within the grade groups is improved if they race only against themselves. It also maintains the suspense, which is important to many. Stagger started racing puts less time between each scouts heats.

More to follow ...

[Stan Pope]

Reply 2 of 2

All that said, I suspect that the organization has some fixed ideas about how the competition should take place, and would be open only to limited changes to accomplish accuracy improvement.

Check first to see that the group really does want accuracy in the awards. Some actually like the idea that a 5th fastest car plus some luck are satisfactory basis for the 3rd place trophy. I'll continue on the assumption that your group would prefer accuracy rather than a lottery.

I understand that the primary goal of the first racing session is to select finalists for the grade finals and, maybe, to select finalists for the "king of the hill" (whole pack) finals.

You could (A) run all the cars against each other or (B) interlace heats from separate grade charts.

A. You don't say how many rounds of Stearns are used... my guess is 6. You could also use a PPN chart. Average accuracy for the two methods are comparable, but PPN will be more consistent, since it has fewer randomly generated anomalies.

In either case, you need to select 6 or 7 finalists to have a high probability that the group of finalists includes the three fastest cars. To pick finalists for subgroup competition, 6 or 7 finalists is probably less certain than for the selection of finalists for the overall competition. I don't have simulations to base this estimate on ... just analysis that error from opponent imbalance is augmented by error from selecting from further down the list of competitors.

B. You could run 6-lane PPN charts for each age group and interlace the heats ... Run a heat of Tigers, a heat of Wolves, a heat of Bears, a heat of 4th grade Webelos, a heat of 5th grade Webelos, another heat of Tigers, etc. This gives you better resolution of the finalists for each group. It also requires a different way of selecting overall finalists. 6 or 7 finalists from each group will have much better probability of including the three fastest cars in each group.... a real plus in my book.

Now, you can take your three most closely matched lanes and run PN 7 or CPN 7 Finals for each group with confidence that you are doing a good job of giving each medal to the Scout who most deserves it.

And, you also have 15 finalists (12 if you don't separate the 4th and 5th grade Webelos) and they can run through a PPN chart to pick 6 or 7 trophy finalists.

Then run those finalists through a PN-7 or CPN-7 chart with confidence that you are doing a good job of giving each trophy to the Scout who most deserves it.

In this way, you have kept all of the Scouts involved in the racing up to the start of finals. and you have kept all of the finalists in the running for trophies until even later.

[gpraceman]

Stan,

It might be helpful to provide a link to your site for info on these charts. I'm not sure if he knows even what PN, PPN, CPN charts are.

[Stan Pope]

Stan,

It might be helpful to provide a link to your site for info on these charts. I'm not sure if he knows even what PN, PPN, CPN charts are.

From the question, I bet Tom does know, but for other newer readers, I'll answer anyway!

PPN, PN, and CPN are varying accuracy Young&Pope charts. A javascript generator process can be used online (or saved for offline use) at http://members.aol.com/standcmr/ppngen.html at no cost.

PPN, PN, and CPN are also licensed in executable form in some race management software such as is supplied at
http://grandprix-software-central.com for a modest cost, but you would have to check (or ask Randy Lisano) to see how well races could be interlaced.

[Jungle Jim]

Since you're interested in a scheduling method, I would also have to suggest a different charting system. The ones Stan shows the links to should suffice.

A question I would ask you, and you may want to ask the race committee, is if the intent is to deduce to the fastest cars in your Pack, the fastest in each rank or both?

[ExtremePWD]

Our pack has about 74 boys in it and we have a 4 lane track with electronic timers. For each age group we are switching from a double elimination (stop laughing) to a PPN to sort out the top 7 finalists who move on to a PN7 to award the trophies. We also traditionally ran a single elimination Best Gas Mileage competition with the slower cars. (This is discussed under a different topic) After we sort out the 7 speed finalists we are taking everyone left over and running a PPN to find the slowest car. Not as accurate as just taking the slowest 7 and running a PN7 but we wanted to maximize racing for all of the scouts. Running the PPN feeding into a PN7 and a residual PPN will result in all boys getting 8 races (not including tie breaks). The double elimination we ran before followed by a single elimination for the slow cars had resulted in some boys going home after only 3 races. (Thanks again Stan for the help)

[PWTom]

Thanks for all of the input. Last year we had approximately 78 boys entered and used an 84 car Stearns that yielded 20 rounds with 14 heats per round. Lots of racing (about 4 hours)! Most people in the pack are convinced because we have so many rounds/heats that we have a highly accurate system

My take on the Pack's approach is that they are more interested in determining the fastest cars at the Pack level, and the places within the ranks fall out from the overall Pack "order."

In using the various chart testing tools, a 20-round Stearns has a 70% accuracy of the top 6 cars, while a 84-6 PPN has an accuracy of 83% (with a lot fewer races!). I am considering running the 84-6 PPN to generate the "fastest 6" from the pack, then run a CPN6-6(12) chart for the finals.

Are there any methods of selecting the final six (or what ever number is appropriate) that would yield a higher accuracy for the finals?? Additional comments on the overall strategy are also appreciated. My hope is to move the Pack towards more accurate methods over the next three years.

[Stan Pope]

In using the various chart testing tools, a 20-round Stearns has a 70% accuracy of the top 6 cars, while a 84-6 PPN has an accuracy of 83% (with a lot fewer races!). I am considering running the 84-6 PPN to generate the "fastest 6" from the pack, then run a CPN6-6(12) chart for the finals.

Sounds like you are reallly doing your homework! Congratulations.

Just to be sure, the quoted numbers were for "3n6", meaning "probability that all three fastest cars are among the 6 highest scoring." I ask because I expected somewhat higher numbers. The numbers quoted look more like "6n6" or, perhaps, "3-trophy" accuracy number.

The "3n6" numbers are appropriate for screening cars into more accurate finals. You don't really care about the finish order within the top 6, just that the top 6 include the 3 objectively fastest cars.

Compare the Top-3 percentages for the PPN-84 and the Stearns to the product of Top-3 percentage for a CP6-6 and the 3n6 percentage for the PPN-84. This is the analysis to base the decision on. You probably find that the latter awards the trophies more accurately.

Are there any methods of selecting the final six (or what ever number is appropriate) that would yield a higher accuracy for the finals?? Additional comments on the overall strategy are also appreciated. My hope is to move the Pack towards more accurate methods over the next three years.

I don't know of any other schemes which do as well unless you give up the requirement that Scouts and Parents can observe the accuracy and fairness of the process. This means that there are no hidden processes and arbitrary decisions.

There are several definitions of "fastest car", because several factors go into the selection.

In a long elimination series where the hardest heats are at the end, the fastest car may be the one that maintained its lubrication effectiveness the longest.

In a timed series, it may be the raw total time for N heats.

Or one might discard some number of fastest and slowest heats.

Or one might discard all but the fastest heat.

In final standings, it might be based on how many cars each has beaten in the course of racing.

Consistency and staying power play different toles in each. So, you hay have to decide first what you mean by "fastest".

[Stan Pope]

In performing the method eveluations, be sure to keep the assumptions between them consistent. You may also want to test the sensitivity of the results to various assumptions by rerunning simulations with bracketing values (what happens if this assumption is -30% and what happens if it is +30%?). That should give you more confidence in the results.

Are you using Cory Young's chart simulation program? (It is the only one that I know of for the purpose.) He did a nice job on it, I think! If not, what?

[PWTom]

Thanks for the education on what some of the outputs mean. It has been too many years since my stats class!

I'll do more case studies on combinations of methods to see which yields the best accuracy while still maintaining lots of racing fun. I have been using Cory Young's PWD Tools. They are terrific. With today's PC power, I can run 200 simulations of a large chart in just a couple of minutes.

[Jungle Jim]

My take on the Pack's approach is that they are more interested in determining the fastest cars at the Pack level, and the places within the ranks fall out from the overall Pack "order."

If you know for a fact that is what they want, then a single, inclusive race is fine. Of course a big flaw is that you are greatly reducing the probability of sending your fastest racers to the next level (if your council has that).

[Stan Pope]

If you know for a fact that is what they want, then a single, inclusive race is fine. Of course a big flaw is that you are greatly reducing the probability of sending your fastest racers to the next level (if your council has that).

Excellent point!

What is the plan in your area for participation in district/council races, Tom?

In my district each pack sends 4 fastest in each grade. In my Grandson's council, each pack sends three fastest overall, without regard to grade. In a neighboring district, anybody who wants to race in the district race may participate. (BTW, can you guess which race has the highest attendance, in spite of drawing from the smallest pool of boys?)

[PWTom]

Our District race is open to all comers (I believe). I am embarassed to say that I know very little about our district race, even though my son has done fairly well in our Pack races during his first two years. Our Pack has not promoted participation in the District (I am hoping to change that).

[Darin McGrew]

When we've had regional derbies, they've been open to all comers as well. In fact, some smaller CSB Stockade units have used the regional derby as their annual derby, rather than putting on their own local derby for a handful of boys.